"Data science" can often be an amorphous and murky world for organizations, no matter the industry or sector. Collecting and having data is one thing, but actually knowing what to do with and what kind of resources to allocate tends to be a mystery for many.
There are, however, many delusions that organizations tend to harbor about data science and its capabilities. Steering clear from these is easier said than done, especially for those who see the long-term potential and value of data itself. But in order to preserve the longevity and really drive success, it absolutely essential to steer clear of certain delusions that are associated with data science.
A recent article expounded on some of the key delusions that organizations face as a whole, but there are some takeaways from the article that can really be applied to more niche sectors such as talent in order to see a different perspective - here are some of the highlights.
Data Delusion #1: Available data is always accurate
When it comes to talent, there are lots of quick assumptions that can be made, especially when it comes "willingness to move jobs" that is based on information that is available publicly. Unfortunately, this is a major stretch, especially for tech-savvy people.
This concept applies to the concept of "diversity prediction" as well. The drive for answers is strong, and that means that it is often very easy to be blinded. This is compounded by the amount of information that is publicly available, but unfortunately, just because it is available does not mean that it will be accurate.
The truth is, the capabilities of publicly available data when it comes to providing accurate screenshots of situations is extremely limited. While it seems like technology is innovating in a way to make that possible, that is currently not the case. Resist the urge to go for what is at face-value, and think of ways to evaluate your current data at a deeper level to really get the results you want.
Data Delusion #2: All you need is a simple data set
Believe it or not, human behavior is complex. When trying to construct likelihoods of hiring, turnover, and performance, it might easy to rely on simple datasets. But think about it, how can simple data predict complex behavior?
That means the process itself is flawed because the datasets will continually struggle to provide good outcomes and lead you to a circle of misused data, inaccurate results and lower predictability. The article makes a strong point about differentiating between simple manual processes to understand data and what aspects of the data need a deeper focus to eliminate this.
Data Delusion #3: There is always an insight hiding in the data
"Overheard: "Just give them a dump of all the data. They will use data science to make the insights pop out""
The article's quote is spot-on in this instance. Companies and vendors are quick to hire data scientists, build complex talent algorithms and other processes...when there isn't any expertise to keep that momentum going.
This can be incredibly problematic in the long-run, and a whole lot of frustrations if you are not quite getting the results you want. In order to avoid this delusion, it is better to work with others that do have the right domain expertise in order to ensure that you are getting value out of your data. Having scalable, defined projects and grounding those in real outcomes will do volumes to ensure that your resources are being used correctly.
Data Delusion #4: Data scientists must fit a predefined box of skills
It might be tempting to heavily focus on certain areas of expertise, such as HR Data Analysts, statisticians or computer scientists when analyzing data. While yes, they have a certain level of skill and expertise that might initially seem necessary for the job, think about the big picture.
Having narrowly defined roles and skill sets might not lead to the best outcomes. Many do not believe this, but diversity in data science can be a meaningful asset that really ensures that you understand the complete picture your data can show you. Another important point to consider is that diversity across teams is especially important when you consider the room for bias and errors that this allows for in both the data and the way the team operates.
Data Delusion #5: AI v. Analyst
There is a constant discussion about whether AI will replace the analyst, or really just a broader debate about whether or not technology has the power to completely outshine the human touch. The truth is, one cannot really survive without the other.
It might be tempting to go all-in on an AI strategy, or go the opposite direction and look for the "unicorn data scientist" as the article puts it, but neither approach will really get you where you need to be. Judging by the paucity of data (relatively speaking) in talent, this shift towards total AI will take longer than in other disciplines.
Data Delusion #6: You get science, they do data and the other one does analysis
It might be tempting to silo off roles and have a full matrix team dedicated to the data. But unfortunately, all that really does is breed inefficiency, and none of the team gets the resources they need to deal with the data. With AI-powered solutions like ours at Swooptalent, we have the technology handle most of the heavy lifting in that regard, so that is one less problem you would have to worry about.
We aren't saying you shouldn't have data science as a part of your talent strategy - you definitely should. But try hard not to wade into too many delusions. Oh, and if you need better DATA for your data science work, please contact us - our data is a glorious smorgasbord for data science!